March 10, 2009

Multiple Channels And Long-Term Value

Here's one for you.

I recently ran a regression model to predict future value, based on the following attributes:
  • Months Since Last Purchase
  • Life-To-Date Demand
  • Life-To-Date Merchandise Divisions Purchased From
  • Life-To-Date Advertising Channels Purchased From
All variables were significant. Life-To-Date Advertising Channels was the least significant variable, with the least impact on future demand. Here's a table with results:

Coeff. T-Statistic
Constant 94.387 27.3
Square Root Of Recency (41.483) (45.8)
LTD Demand 0.103 127.7
Square Root Of Merch Divisions 25.972 25.2
Square Root Of Advertising Channels 5.077 5.6

Future Value, $500 6 Month Buyer

Merch = 1, Ad Chan = 1 $75.32
Merch = 2, Ad Chan = 1 $86.08
Merch = 3, Ad Chan = 1 $94.34
Merch = 1, Ad Chan = 2 $77.43
Merch = 2, Ad Chan = 2 $88.18
Merch = 3, Ad Chan = 2 $96.44
Merch = 1, Ad Chan = 3 $79.04
Merch = 2, Ad Chan = 3 $89.80
Merch = 3, Ad Chan = 3 $98.05

Notice that each additional merchandise division a customer purchases from adds more future value than does each additional advertising channel (catalogs, e-mail, paid search, affiliates, portal advertising, etc).

The lesson here isn't that multiple channels are bad. They're good!

The lesson is to think about channels as a transition-based attribute, and not as a customer-value attribute.

Think about your own behavior.

In 1989, you phoned a call center, ordered merchandise from a catalog, and waited 4-6 weeks for it to arrive.

In 1999, you received a catalog, then keyed your order in to a website.

In 2002, you received a catalog and an e-mail campaign, then keyed your merchandise into a website.

In 2004, you searched Google for competing products, then used the search feature on the website to place an order.

In 2007, you were caught up in the buzz about a product on a blog, then purchased on the website.

In 2009, maybe you were caught up in the buzz about a product on Twitter, then purchased on the website.

None of these channels make you a better or worse customer. They simply represent your transition through the maze of evolving technology. It turns out that good customers move through the transition faster than marginal customers.

If you're a member of the research community, selling $1,999 reports, please help us out. The "multichannel customers are worth 3x - 10x more than single channel customers, and make $100,000 a year" comments are holding back our industry.

Honestly, if this logic held, wouldn't Netflix be on the verge of bankruptcy while Blockbuster thrived?

Wouldn't Abe's of Maine be closing, not Circuit City?

Wouldn't be struggling, not Borders?